TEST(UtilTest, HtpParseUri) {
bstr *input = NULL;
htp_uri_t *uri = NULL;
uri_test *test;
input = bstr_dup_c("");
EXPECT_EQ(HTP_OK, htp_parse_uri(input, &uri));
bstr_free(input);
free_htp_uri_t(&uri);
test = uri_tests;
while (test->uri != NULL) {
input = bstr_dup_c(test->uri);
EXPECT_EQ(HTP_OK, htp_parse_uri(input, &uri));
EXPECT_PRED_FORMAT2(UriIsExpected, test->expected, uri)
<< "Failed URI = " << test->uri << std::endl;
bstr_free(input);
free_htp_uri_t(&uri);
++test;
}
}
htp_status_t htp_tx_state_request_line(htp_tx_t *tx) {
if (tx == NULL) return HTP_ERROR;
// Determine how to process the request URI.
if (tx->request_method_number == HTP_M_CONNECT) {
// When CONNECT is used, the request URI contains an authority string.
if (htp_parse_uri_hostport(tx->connp, tx->request_uri, tx->parsed_uri_raw) != HTP_OK) {
return HTP_ERROR;
}
} else {
// Parse the request URI into htp_tx_t::parsed_uri_raw.
if (htp_parse_uri(tx->request_uri, &(tx->parsed_uri_raw)) != HTP_OK) {
return HTP_ERROR;
}
}
// Build htp_tx_t::parsed_uri, but only if it was not explicitly set already.
if (tx->parsed_uri == NULL) {
tx->parsed_uri = htp_uri_alloc();
if (tx->parsed_uri == NULL) return HTP_ERROR;
// Keep the original URI components, but create a copy which we can normalize and use internally.
if (htp_normalize_parsed_uri(tx->connp, tx->parsed_uri_raw, tx->parsed_uri) != HTP_OK) {
return HTP_ERROR;
}
}
// Check parsed_uri hostname.
if (tx->parsed_uri->hostname != NULL) {
if (htp_validate_hostname(tx->parsed_uri->hostname) == 0) {
tx->flags |= HTP_HOSTU_INVALID;
}
}
// Run hook REQUEST_URI_NORMALIZE.
htp_status_t rc = htp_hook_run_all(tx->connp->cfg->hook_request_uri_normalize, tx);
if (rc != HTP_OK) return rc;
// Run hook REQUEST_LINE.
rc = htp_hook_run_all(tx->connp->cfg->hook_request_line, tx);
if (rc != HTP_OK) return rc;
// Move on to the next phase.
tx->connp->in_state = htp_connp_REQ_PROTOCOL;
return HTP_OK;
}
// We return a HTP::URI and throw an exception on error.
VALUE rbhtp_parse_uri( VALUE self, VALUE input )
{
Check_Type( input, T_STRING );
bstr* input_b = bstr_dup_mem( RSTRING_PTR( input ), RSTRING_LEN( input ) );
htp_uri_t* uri = NULL; // htp_parse_uri will alloc.
int result = htp_parse_uri( input_b, &uri );
if ( result != HTP_OK ) {
bstr_free( &input_b );
free( uri );
rb_raise( rb_eRuntimeError, "HTP error in htp_parse_uri: %d", result );
return Qnil; // Ignored?
}
bstr_free( &input_b ); // Okay, as htp_parse_uri dups the data it needs.
return rb_funcall( cURI, rb_intern( "new" ), 1,
Data_Wrap_Struct( rb_cObject, 0, rbhtp_free_uri, uri )
);
}
htp_status_t htp_tx_state_request_line(htp_tx_t *tx) {
htp_connp_t *connp = tx->connp;
if (connp->in_tx->request_method_number == HTP_M_CONNECT) {
// Parse authority
if (htp_parse_uri_hostport(connp, connp->in_tx->request_uri, &(connp->in_tx->parsed_uri_incomplete)) != HTP_OK) {
// Note: downstream responsible for error logging.
return HTP_ERROR;
}
} else {
// Parse the request URI
if (htp_parse_uri(connp->in_tx->request_uri, &(connp->in_tx->parsed_uri_incomplete)) != HTP_OK) {
// Note: downstream responsible for error logging.
return HTP_ERROR;
}
// Keep the original URI components, but
// create a copy which we can normalize and use internally.
if (htp_normalize_parsed_uri(connp, connp->in_tx->parsed_uri_incomplete, connp->in_tx->parsed_uri) != HTP_OK) {
// Note: downstream responsible for error logging.
return HTP_ERROR;
}
// Run hook REQUEST_URI_NORMALIZE.
int rc = htp_hook_run_all(connp->cfg->hook_request_uri_normalize, connp);
if (rc != HTP_OK) return rc;
// Now is a good time to generate request_uri_normalized, before we finalize
// parsed_uri (and lose the information which parts were provided in the request and
// which parts we added).
if (connp->cfg->generate_request_uri_normalized) {
connp->in_tx->request_uri_normalized = htp_unparse_uri_noencode(connp->in_tx->parsed_uri);
if (connp->in_tx->request_uri_normalized == NULL) return HTP_ERROR;
#ifdef HTP_DEBUG
fprint_raw_data(stderr, "request_uri_normalized",
(unsigned char *) bstr_ptr(connp->in_tx->request_uri_normalized),
bstr_len(connp->in_tx->request_uri_normalized));
#endif
}
// Finalize parsed_uri.
// Scheme.
if (connp->in_tx->parsed_uri->scheme != NULL) {
if (bstr_cmp_c(connp->in_tx->parsed_uri->scheme, "http") != 0) {
// TODO Invalid scheme.
}
} else {
connp->in_tx->parsed_uri->scheme = bstr_dup_c("http");
if (connp->in_tx->parsed_uri->scheme == NULL) {
return HTP_ERROR;
}
}
// Path.
if (connp->in_tx->parsed_uri->path == NULL) {
connp->in_tx->parsed_uri->path = bstr_dup_c("/");
if (connp->in_tx->parsed_uri->path == NULL) {
return HTP_ERROR;
}
}
}
// Run hook REQUEST_LINE.
int rc = htp_hook_run_all(connp->cfg->hook_request_line, connp);
if (rc != HTP_OK) return rc;
// Move on to the next phase.
connp->in_state = htp_connp_REQ_PROTOCOL;
return HTP_OK;
}
/**
* Parses request line.
*
* @param connp
* @returns HTP_OK on state change, HTTP_ERROR on error, or HTP_DATA when more data is needed.
*/
int htp_connp_REQ_LINE(htp_connp_t *connp) {
for (;;) {
// Get one byte
IN_COPY_BYTE_OR_RETURN(connp);
// Keep track of NUL bytes
if (connp->in_next_byte == 0) {
// Remember how many NULs there were
connp->in_tx->request_line_nul++;
// Store the offset of the first NUL byte
if (connp->in_tx->request_line_nul_offset == -1) {
connp->in_tx->request_line_nul_offset = connp->in_line_len;
}
}
// Have we reached the end of the line?
if (connp->in_next_byte == LF) {
#ifdef HTP_DEBUG
fprint_raw_data(stderr, __FUNCTION__, connp->in_line, connp->in_line_len);
#endif
// Is this a line that should be ignored?
if (htp_connp_is_line_ignorable(connp, connp->in_line, connp->in_line_len)) {
// We have an empty/whitespace line, which we'll note, ignore and move on
connp->in_tx->request_ignored_lines++;
// TODO How many empty lines are we willing to accept?
// Start again
connp->in_line_len = 0;
return HTP_OK;
}
// Process request line
connp->in_tx->request_line_raw = bstr_dup_mem((char *) connp->in_line, connp->in_line_len);
if (connp->in_tx->request_line_raw == NULL) {
return HTP_ERROR;
}
/// @todo Would be nice to reference request_line_raw data
htp_chomp(connp->in_line, &connp->in_line_len);
connp->in_tx->request_line = bstr_dup_ex(connp->in_tx->request_line_raw, 0, connp->in_line_len);
if (connp->in_tx->request_line == NULL) {
return HTP_ERROR;
}
// Parse request line
if (connp->cfg->parse_request_line(connp) != HTP_OK) {
// Note: downstream responsible for error logging
return HTP_ERROR;
}
if (connp->in_tx->request_method_number == M_CONNECT) {
// Parse authority
if (htp_parse_authority(connp, connp->in_tx->request_uri, &(connp->in_tx->parsed_uri_incomplete)) != HTP_OK) {
// Note: downstream responsible for error logging
return HTP_ERROR;
}
} else {
// Parse the request URI
if (htp_parse_uri(connp->in_tx->request_uri, &(connp->in_tx->parsed_uri_incomplete)) != HTP_OK) {
// Note: downstream responsible for error logging
return HTP_ERROR;
}
// Keep the original URI components, but
// create a copy which we can normalize and use internally
if (htp_normalize_parsed_uri(connp, connp->in_tx->parsed_uri_incomplete, connp->in_tx->parsed_uri)) {
// Note: downstream responsible for error logging
return HTP_ERROR;
}
// Run hook REQUEST_URI_NORMALIZE
int rc = hook_run_all(connp->cfg->hook_request_uri_normalize, connp);
if (rc != HOOK_OK) {
switch (rc) {
case HOOK_STOP:
return HTP_STOP;
case HOOK_ERROR:
case HOOK_DECLINED:
default:
htp_log(connp, HTP_LOG_MARK, HTP_LOG_ERROR, 0,
"Request headers callback returned error (%d)", rc);
return HTP_ERROR;
}
}
// Now is a good time to generate request_uri_normalized, before we finalize
// parsed_uri (and lose the information which parts were provided in the request and
// which parts we added).
if (connp->cfg->generate_request_uri_normalized) {
connp->in_tx->request_uri_normalized = htp_unparse_uri_noencode(connp->in_tx->parsed_uri);
if (connp->in_tx->request_uri_normalized == NULL) {
// There's no sense in logging anything on a memory allocation failure
return HTP_ERROR;
}
#ifdef HTP_DEBUG
fprint_raw_data(stderr, "request_uri_normalized",
(unsigned char *) bstr_ptr(connp->in_tx->request_uri_normalized),
bstr_len(connp->in_tx->request_uri_normalized));
#endif
}
// Finalize parsed_uri
// Scheme
if (connp->in_tx->parsed_uri->scheme != NULL) {
if (bstr_cmp_c(connp->in_tx->parsed_uri->scheme, "http") != 0) {
// TODO Invalid scheme
}
} else {
connp->in_tx->parsed_uri->scheme = bstr_dup_c("http");
if (connp->in_tx->parsed_uri->scheme == NULL) {
return HTP_ERROR;
}
}
// Port
if (connp->in_tx->parsed_uri->port != NULL) {
if (connp->in_tx->parsed_uri->port_number != -1) {
// Check that the port in the URI is the same
// as the port on which the client is talking
// to the server
if (connp->conn->use_local_port) {
if (connp->in_tx->parsed_uri->port_number != connp->conn->local_port) {
// Incorrect port; use the real port instead
connp->in_tx->parsed_uri->port_number = connp->conn->local_port;
// TODO Log
}
} else {
connp->in_tx->parsed_uri->port_number = connp->conn->remote_port;
}
} else {
// Invalid port; use the real port instead
if (connp->conn->use_local_port) {
connp->in_tx->parsed_uri->port_number = connp->conn->local_port;
} else {
connp->in_tx->parsed_uri->port_number = connp->conn->remote_port;
}
// TODO Log
}
} else {
if (connp->conn->use_local_port) {
connp->in_tx->parsed_uri->port_number = connp->conn->local_port;
} else {
connp->in_tx->parsed_uri->port_number = connp->conn->remote_port;
}
}
// Path
if (connp->in_tx->parsed_uri->path == NULL) {
connp->in_tx->parsed_uri->path = bstr_dup_c("/");
if (connp->in_tx->parsed_uri->path == NULL) {
return HTP_ERROR;
}
}
}
// Run hook REQUEST_LINE
int rc = hook_run_all(connp->cfg->hook_request_line, connp);
if (rc != HOOK_OK) {
switch (rc) {
case HOOK_STOP:
return HTP_STOP;
case HOOK_ERROR:
case HOOK_DECLINED:
default:
htp_log(connp, HTP_LOG_MARK, HTP_LOG_ERROR, 0,
"Request headers callback returned error (%d)", rc);
return HTP_ERROR;
}
}
// Clean up.
connp->in_line_len = 0;
// Move on to the next phase.
connp->in_state = htp_connp_REQ_PROTOCOL;
return HTP_OK;
}
}
}